Posts

“Apertus: Democratizing Open and Compliant LLMs for Global Language Environments” – a project by Team Apertus within the Swiss AI Initiative.

“Benchmarking Optimizers for Large Language Model Pretraining” – joint work with Matteo Pagliardini and Martin Jaggi.

“Gradient-Normalized Smoothness for Optimization with Approximate Hessians” – joint work with Martin Jaggi and Nikita Doikov.

“Clipping Improves Adam-Norm and AdaGrad-Norm when the Noise Is Heavy-Tailed” (with Savelii Chezhegov, Yaroslav Klyukin, Aleksandr Beznosikov, Alexander Gasnikov, Samuel Horváth, Martin Takáč and Eduard Gorbunov).

“Sign Operator for Coping with Heavy-Tailed Noise: High Probability Convergence Bounds with Extensions to Distributed Optimization and Comparison Oracle” – joint work with Nikita Kornilov, Philip Zmushko, Alexander Gasnikov and Aleksandr Beznosikov.

“Just a Simple Transformation is Enough for Data Protection in Vertical Federated Learning” – joint work with Philip Zmushko, Alexander Pichugin and Aleksandr Beznosikov.